Research On Task Offloading Strategy For Multi-MEC Collaboration System

With the rapid development of mobile telecommunications technology and smart terminals,a large number of new computing-intensive and delay-sensitive applications continue to emerge,such as smart transportation,virtual reality,Internet of Vehicles,Internet of Things,etc.,in order to meet the low-latency response of mobile terminals Demand,a distributed computing paradigm that integrates heterogeneous resources at the edge of the network is proposed,namely Multi-access Edge Computing(MEC),also known as mobile edge computing.However,due to the limited computing resources of the MEC and the complex and diverse task types of the user terminal,there is a large gap in the processing difficulty of different task types.Therefore,the offloading strategies of different task types in the multi-MEC scenario still need to be further studied.In response to the above problems,this article has carried out the following research:1.Aiming at the computational offloading problem in the user aggregation scenario of multi-user and multi-Fog-Access Point,it is divided into two steps to deal with.First,an FAP selection algorithm under multiple signal coverage is proposed to select a suitable F-AP for each terminal and reduce the F-AP load of the central layer to achieve the effect of load balancing.Secondly,to improve the service experience quality of the central layer task offloading request,a hierarchical collaborative heuristic area computing offloading algorithm is designed to abstract the idle computing resources of the sparse layer and the transition layer into higher-level processing units,and at the same time,the tasks overflowing from the central layer are scheduled to the higher-level F-AP,thereby improving the success rate of task calculation.At the same time,under the premise of ensuring that each offloading task can be completed within a limited time,a lower level of computing resources is allocated.After all tasks are guaranteed to be processed,the level of allocated resources is gradually increased within the range of available computing resources to reduce task offloading total delay.The experimental results show that the algorithm has good offloading performance in user gathering scenarios.2.Aiming at the computing offloading scenario of multiple dependent tasks in the single-user multi-F-AP scenario,in order to reduce the complexity of dependent task processing,the dependency relationship is abstracted into a directed acyclic graph(DAG),and the edges represent the mandatory dependencies between tasks,and the nodes represent subtasks.In order to prevent the subsequent task from waiting too long for the result of the predecessor task during the task uninstallation process,it should be uninstalled in order according to the dependency of the DAG.Here,the upstream weight traversed from the exit task is used as the priority of the task,and then sorting in descending order according to the priority can obtain the offloading sequence.In order to reduce the completion time of dependent tasks,a scheduling method that makes full use of the idle time slots of the computing unit and a list scheduling algorithm for copying key tasks are proposed.The experimental results prove that the algorithm has better performance than other algorithms in the current scenario.